scholarly journals Modeling Recidivism through Bayesian Regression Models and Deep Neural Networks

Mathematics ◽  
2021 ◽  
Vol 9 (6) ◽  
pp. 639
Author(s):  
Rolando de la Cruz ◽  
Oslando Padilla ◽  
Mauricio A. Valle ◽  
Gonzalo A. Ruz

This study aims to analyze and explore criminal recidivism with different modeling strategies: one based on an explanation of the phenomenon and another based on a prediction task. We compared three common statistical approaches for modeling recidivism: the logistic regression model, the Cox regression model, and the cure rate model. The parameters of these models were estimated from a Bayesian point of view. Additionally, for prediction purposes, we compared the Cox proportional model, a random survival forest, and a deep neural network. To conduct this study, we used a real dataset that corresponds to a cohort of individuals which consisted of men convicted of sexual crimes against women in 1973 in England and Wales. The results show that the logistic regression model tends to give more precise estimations of the probabilities of recidivism both globally and with the subgroups considered, but at the expense of running a model for each moment of the time that is of interest. The cure rate model with a relatively simple distribution, such as Weibull, provides acceptable estimations, and these tend to be better with longer follow-up periods. The Cox regression model can provide the most biased estimations with certain subgroups. The prediction results show the deep neural network’s superiority compared to the Cox proportional model and the random survival forest.

2008 ◽  
Vol 23 (4) ◽  
pp. 251-259 ◽  
Author(s):  
Theodora Bejan-Angoulvant ◽  
Anne-Marie Bouvier ◽  
Nadine Bossard ◽  
Aurelien Belot ◽  
Valérie Jooste ◽  
...  

2018 ◽  
Vol 87 (5) ◽  
pp. 255-262 ◽  
Author(s):  
A. Dufourni ◽  
A. Decloedt ◽  
L. Lefère ◽  
D. De Clercq ◽  
P. Deprez ◽  
...  

While mature coastal bermudagrass hay is strongly associated with ileal impaction in the Southeastern United States, stabling on flax bedding has anecdotally been associated with this condition in Europe. The aim of this retrospective study was to investigate the association between ileal impaction and the use of flax shives compared to straw as bedding in horses with colic. Medical records of 2336 referral cases evaluated for abdominal pain between January 2008 and May 2017 at the Department of Large Animal Internal Medicine, Ghent University were reviewed. Diagnosis, date of admission, age, breed, gender, body weight and stable bedding were recorded. Conditional logistic regression analysis was used to assess the association between ileal impaction and each individual variable. Odds ratios (OR) and 95% confidence intervals (CI) were determined. Predictors with a value of P < 0.2 were included in a multivariable Cox regression model and Wald’s test was used to assess parameter estimate significance. Further, the association between survival to discharge and type of bedding or type of treatment (medical versus surgical) was analyzed for horses with ileal impactions. The proportion of colic cases stabled on flax bedding at home was 11.3%. The overall prevalence of ileal impaction was 4.2%. In the flax group, the prevalence of ileal impaction was 9.4% as opposed to 3.6% within the straw group. The OR of 2.8 (95% CI 1.7-4.7; P < 0.001) in the multivariable logistic regression model indicated that horses stabled on flax shives were approximately three times more likely to have ileal impactions than horses stabled on straw. There was no significant association found between ileal impaction and the period of admission, age, gender or body weight in a multivariable logistic regression model. The odds for having ileal impaction is approximately six times (OR 6.3; 95% CI 2.4-16.4; P < 0.001) higher in draft horses than in warmbloods in the multivariable logistic regression model. No significant association was found between survival to discharge and type of bedding or treatment. These results suggest that horses with colic that were housed on flax bedding are more likely to present ileal impactions than horses housed on straw.


Author(s):  
Yangmin Hu ◽  
Danyang Li ◽  
Lingcheng Xu ◽  
Yuping Hu ◽  
Yiwen Sang ◽  
...  

Abstract Background Infection is the leading cause of morbidity and mortality among burn patients, and bloodstream infection (BSI) is the most serious. This study aimed to evaluate the epidemiology and clinical outcomes of BSI in severe burn patients. Methods Clinical variables of all patients admitted with severe burns (≥ 20% total body surface area, %TBSA) were analyzed retrospectively from January 2013 to December 2018 at a teaching hospital. The Kaplan–Meier method was utilized for plotting survival curves. Multivariate logistic regression and Cox regression model were also performed. Results A total of 495 patients were evaluated, of whom 136 (27.5%) had a BSI. The median time from the patients being burned to BSI was 8 days. For BSI onset in these patients, 47.8% (65/136) occurred in the first week. The most frequently isolated causative organism was A. baumannii (22.7%), followed by methicillin-resistant Staphylococcus aureus (18.7%) and K. pneumoniae (18.2%), in patients with BSI. Multivariate logistic regression analysis showed that %TBSA (p = 0.023), mechanical ventilation (p = 0.019), central venous catheter (CVC) (p < 0.001) and hospital length of stay (27d vs 50d, p < 0.001) were independent risk factors associated with BSI. Cox regression model showed that acute kidney injury (HR, 12.26; 95% CI 2.31–64.98; p = 0.003) and septic shock (HR, 4.36; 95% CI 1.16–16.34; p = 0.031) were identified as independent predictors of 30-day mortality of BSI in burn patients. Conclusions Multidrug resistant gram-negative bacteria were the main pathogens of BSI in severe burn patients. Accurate evaluation of risk factors for BSI and the mortality of BSI in severe burn patients may improve early appropriate management.


Sign in / Sign up

Export Citation Format

Share Document